Ecology, 84(10), 2003, pp. 2589±2599 q 2003 by the Ecological Society of America SPATIAL ECOLOGY OF PREDATOR±PREY INTERACTIONS: CORRIDORS AND PATCH SHAPE INFLUENCE SEED PREDATION JOHN L. ORROCK,1,4 BRENT J. DANIELSON,1 MOLLY J. BURNS,2 AND DOUGLAS J. LEVEY3 1Ecology and Evolutionary Biology Interdepartmental Graduate Program, 124 Science II, Iowa State University, Ames, Iowa 50011-3221 USA 2Department of Natural Resource Ecology and Management, 124 Science II, Iowa State University, Ames, Iowa 50011-3221 USA 3Department of Zoology, University of Florida, Gainesville, Florida 32611-8525 USA Abstract. Corridors that connect patches of disjunct habitat may be promising tools for mediating the negative impacts of habitat fragmentation, but little is known about how corridors affect ecological interactions. In eight 12-ha experimental landscapes, we ex- amined how corridors affect the impact of invertebrate, rodent, and avian seed predators on pokeweed, Phytolacca americana. Over 13 months in 2000 and 2001, we quanti®ed the effects of patch shape, connectivity, and predator type on the number of seeds germinating in the ®eld (germinants), seed removal, and the viability of remaining seeds. Corridors did not affect the number of P. americana germinants in experimental ex- closures or the viability of seeds remaining in exclosures. However, corridors affected the removal of seeds in a predator-speci®c manner: invertebrates removed more seeds in un- connected patches, whereas rodents removed more seeds in connected patches. Seed removal by birds was similar in connected and unconnected patches. Total seed removal by all seed predators was not affected by corridors, because invertebrates removed more seeds where rodents removed fewer seeds, and vice versa. Overall, seed predation signi®cantly reduced the number and viability of remaining seeds, and reduced the number of germinants in 2000 but not in 2001. The abundance of naturally occurring P. americana plants in our experimental patches in 2000 decreased with increasing seed removal from exclosures but was not related to viability or germinants in 2000, suggesting that seed removal may shape the distribution and abundance of this species. Complementary patterns of seed removal by rodents and invertebrates suggest that corridors alter the effects of these predator taxa by changing the relative amounts of edge and core (nonedge) habitats in a patch. Because invertebrates and rodents do not completely overlap in the seeds they consume, corridors may change predation pressure on seeds that are primarily consumed by one predator type, with potential consequences for the com- position of plant and seed predator communities. Key words: corridors; patch shape; Phytolacca americana; predator±prey interactions; Savannah River Site, South Carolina (USA); seed predation. INTRODUCTION serve a target species only to discover that an important predator uses corridors more effectively than the spe- Habitat fragmentation and species loss are occurring cies of concern. on a global scale, yet few experimental studies have By facilitating predator movement, corridors may im- examined the effects of fragmentation on ecological pact prey that rely upon ``predator-free'' space (e.g., Holt communities (Gonzalez et al. 1998, Collinge 2000, De- and Lawton 1993). Moreover, corridors could alter the binski and Holt 2000, Davies et al. 2001). Corridors spatial or temporal asynchrony that promotes the per- connect habitat fragments and are thought to promote sistence of some predator±prey interactions (e.g., Earn population persistence by promoting gene ¯ow, pop- et al. 2000). Empirical work in microcosms reveals that ulation rescue, and increasing abundance (Rosenberg corridors can affect persistence of predator±prey systems et al. 1997). However, corridors have been criticized (Holyoak and Lawler 1996), but also warns that con- because mechanisms underlying purported corridor ef- necting patches can sometimes lead to counterintuitive fects are unknown (Simberloff et al. 1992). Perhaps outcomes (Burkey 1997, Holyoak 2000). We use an ex- more worrisome, the population-level focus of most perimental landscape to evaluate how corridors affect corridor studies neglects the rest of the ecological mi- the ecology of predator±prey interactions. lieu, with largely unknown consequences. For example, imagine the impact of constructing corridors to pre- Study system: corridors, seeds and seed predators Manuscript received 22 July 2002; revised 24 January 2003; Seeds can be considered sedentary prey consumed accepted 3 February 2003. Corresponding Editor: E. S. Menges. by vertebrate and invertebrate predators (Janzen 1971, 4 E-mail: [email protected] Hulme 1998). Seed predation can lead to dramatic re- 2589 2590 JOHN L. ORROCK ET AL. Ecology, Vol. 84, No. 10 ductions in the standing crop of seeds and may be the primary determinant of plant distribution in some sys- tems (e.g., Louda 1989, Brown and Heske 1990, Hulme 1998, Howe and Brown 1999, 2000, 2001). By in- creasing the deposition of bird-dispersed seeds (Tewks- bury et al. 2002, Haddad et al. 2003), corridors could have positive impacts on bird-dispersed plants. How- ever, the ultimate effect of corridors on bird-dispersed plants could be negative if corridor-mediated changes in seed predation offset or outweigh corridor-mediated increases in seed deposition. Corridors may affect seed predation by altering the movement of seed predators between connected patch- es of suitable habitat (``corridor effects'') or by making it more likely that an individual moving through an uninhabitable matrix will encounter a suitable patch (``drift-fence effects'' [Rosenberg et al. 1997]). Cor- ridors also add area to a patch, and may affect seed predation by providing more habitat for seed predators (``area effects'' [Haddad and Baum 1999]). Moreover, FIG. 1. Layout of the experimental landscape at the Sa- because corridors tend to be relatively long, linear el- vannah River Site, South Carolina. Eight experimental units were created by clearing mature pine forest. Each experi- ements, they can induce area effects by changing the mental unit consisted of connected (C) and unconnected amount of edge habitat relative to the amount of core patches that were either rectangular (R) or winged (W). All habitat in a patch. Mammals (Bennett 1990, La Polla patches were clearcut habitats separated by 150 m of mature and Barrett 1993, Bennett et al. 1994, Bowne et al. pine forest. Within each patch, one of each exclosure type 1999, Coffman et al. 2001, Haddad et al. 2003), in- was installed around the center of the patch (®nal inset). vertebrates (Haddad 1999, 2000, Haddad and Baum 1999, Collinge 2000), and frugivorous birds respond to corridors (Tewksbury et al. 2002). By affecting the ration of our study allowed us to integrate temporal movement and abundance of a particular type of seed variation in seed predation (Whelan et al. 1991) not predator, corridors could change seed predation by that incorporated in many seed predation studies. (Most predator type. studies last ,4 wk, e.g., Hyatt 1998.) Moreover, the Pokeweed, Phytolacca americana, is a perennial duration of our experiment allowed us to evaluate the plant indigenous to eastern North America that is typ- importance of corridors, patch shape, and seed pre- ically found in early successional habitats (Mitich dation during a critical period of establishment for P. 1994). Phytolacca americana produces berries con- americana in our study system. sumed by many bird species (Mitich 1994), which sub- Our objectives are framed as three questions: (1) Do sequently disperse P. americana seeds (McDonnell et corridors affect seed predation by invertebrates, ro- al. 1984, Mitich 1994). Pokeweed seeds are 2.5±3 mm dents, and birds? (2) Are rates of overall seed predation in size (Radford et al. 1968) and can remain viable in affected by corridors? (3) Is the abundance of naturally the seed bank for at least 40 years (Toole and Brown occurring P. americana plants in our study system re- 1946). Predispersal seed predation is probably mini- lated to rates of seed predation, i.e., is there evidence mal, as rodents reject pokeweed fruits, but readily con- that seed predators affect the distribution of P. amer- sume P. americana seeds (McDonnell et al. 1984, Hyatt icana? 1998), and damage to P. americana fruits by rodents METHODS or invertebrates was never observed during ®eld col- lection of several thousand fruits (J. L. Orrock et al. Experimental design personal observation). Thus, if corridors affect seed Experimental landscape.ÐThe experiment was con- predators, there may be direct consequences for P. ducted in eight replicated experimental units created at americana population dynamics. the Savannah River Site, a National Environmental Re- We used an experimental landscape composed of search Park (NERP) near Aiken, South Carolina. Each clearcut patches separated by a pine forest matrix to experimental unit consisted of ®ve patches created dur- evaluate the effect of corridors on seed predators and ing the fall and winter of 1999 by clear-cutting mature Phytolacca seeds. For 13 months, we measured three pine forest (the matrix) followed by prescribed burning different metrics to determine the impact of seed pred- (Fig. 1). In each experimental unit, there were three ators on P. americana: the number of seedlings emerg- different patch types: connected, rectangular, and ing (hereafter called germinants), the number of seeds winged (Fig. 1).